Emotions,Beliefs,and Revisions

Affects and Their Dynamics

Understanding affects as interactions between internal and external dynamics does not only take into account the specifity of affects. It may also solve the problem raised by “steady state” emotions: emotions triggered by situations in keeping with our present expectancies (Frijda, 1993). An example would be our feeling of continuous happiness while appreciating over and over again a landscape the beauty of which meets our expectancies. This may be true in the short term; nevertheless our emotion would decrease or change if we did not notice different features from time to time while staring at the landscape. This suggests that the internal dynamics of happiness can decrease when no longer supplied by different features. For our happiness to continue, our own internal dynamics has to be sustained by the external dynamics offered by the appearance of these different features. Only very essential internal dynamics, related for example to our motivations for survival, for alternating activity and rest, or for love and friendship, can be active almost without external incentives. Even the desires strengthened by the absence of the desired object require signs of this specific absence. Considering affect as an interaction between internal and external dynamics is not only a condition for combining the concepts of revision and emotion, it is also an answer to the “steady state” objection.

The internal dynamics is intentional (Frijda, 1986, 1993; Frijda, Manstaed, & Bem, 2000) in the philosophical sense: we are sensitive to a situation under a given aspect, and as intentionality implies the possibility of misperception, this aspect is submitted to conditions of correction. The affect-as-information theory (Bless, Bonher, Schwarz, & Strack, 1990; Clore & Huntsinger, 2007, 2009; Schwarz, 1990; Schwarz & Clore, 1983; for an evaluation of cognitive theories of emotions, see Moors, 2007) has the same implication (see also Albarracin & Kumkale, 2003; Schwarz & Clore, 1983). The constituents of the dynamics—our desires, beliefs, motivations, goals, and expectancies—define the intentional aspect. Viewing the external dynamics (movement, intensity of light and color, salient forms, facial expressions) as intentional could seem to be an illusory projection of our intentionality because the external dynamics is appraised relative to the orientations of the internal dynamics. For example, tourists enjoy sunny weather while farmers are anxious about drought. Although the sun does not shine intentionally, however, both relations between the sunny weather and these two intentional aspects are a reality. Human facial expressions, animal movements, and the good or bad conditions of living organisms (including plants), all acquire intentionality because by changing their relations with us, they acquire the capacity to change our attitudes. More generally, if internal dynamics can lead to the exploration of possibilities other than the present ones, external dynamics impose real constraints on their exploitation (leading to the exploration–exploitation trade-off, see e.g., Tokic & Günther, 2011). Affects are intentional interactions between our internal dynamics and the external ones.

I disagree with Reisenzein (2009a, 2012) on this point. He claims that the intentionality and object-directedness of affect is an illusion and that affect is only the effect of a cause, which is a combination of desires and revised beliefs. This would imply that affect in the emotion is posterior to the cognitive-conative combination of desire and beliefs. However, we know that the emotional limbic circuit can make us react to a stimulus in a way that is relevant to our motivations before it is processed by cortical and more cognitively elaborated circuits, related to beliefs. We also know that an effect cannot come before its cause (at least at this scale). Moreover Reisenzein’s theory seems to imply that as soon as a desire has been fulfilled and the belief about this fulfillment updated, the cause of the emotion (e.g., happiness) is no longer active. Explaining the persistence of happiness by suggesting that the effect can last longer than its cause is a rather ad hoc solution, particularly if affect comes before the explicit belief that is supposed to be its cause. All these considerations favor a conception of affect as an intentional interaction between internal and external dynamics. Deonna and Teroni (2012) were more on the right track inasmuch as they viewed affects as resulting from the interaction of bodily feelings and evaluative attitudes. This combination implies the intentionality of affects. However, this does not go far enough. We could have bodily feelings and evaluative attitudes, but still feel no emotion. Emotion requires us to be moved by the impact that a dynamics that usually comes from the outside world has on our motivational dynamics—without excluding the fact that a dynamics can also be called “external” by extension in the sense that it has different sources from that of the present motivational flux (e.g., memories or new bodily troubles).

This could be challenged on the grounds that the notion of an internal dynamics tending to slow down, if not strengthened by the external dynamics could be reminiscent of Carver and Scheier’s theory (1998), which implies only internal dynamics. For them, a primary dynamics is linked to our expectancy of satisfying a goal and to a discrepancy with this expectancy. A secondary dynamics is linked to our attempts to reduce the discrepancy. If the ratio or speed of this reduction is higher (or lower) than a reference value, we feel emotions. However, steady emotions also pose a problem for this theory. When the reduction of the discrepancy has appeared to be faster than expected, we will revise the reference value upwards. The continuation of a pleasant situation will therefore no longer raise happiness because the continuation is no longer above the revised reference value.

Here again, steady emotions raise a problem, and we need the external dynamics to compensate for the weakening of the internal one. In addition, the interaction of the two dynamics makes setting a reference value useless. If, for example, the impact of the external dynamics strengthens the internal dynamics, the speed of the former increases the speed of the latter (metaphorically speaking). Therefore, no reference value is required; only the appraisal of the acceleration.

According to Carver and Scheier’s (1998) theory, the intensity of recurrent positive emotions (in similar situations separated by rather long intervals) is supposed to decrease. However, we can enjoy listening to our favorite piece of music from time to time with the same pleasure. When we no longer hear the music for a while, the external and internal dynamics decrease, as our memory of the piece of music is less rich in detail than the real sounds and their external dynamics. Our motivation for hearing real music again will increase with time. When we listen to the piece of music again, this internal dynamics is restrengthened by the richer external dynamics, and we feel the same pleasure. Nevertheless, if we hear another exceptional piece of music, the comparison between the two experiences may decrease our previous pleasure. Both facts are again explained by the difference in strength of two different interactions between internal and external dynamics.

The repetition of the negative emotion of this disappointing comparison between the increasing dynamics of this new exceptional experience and the stationarity of the older one will be an incentive to revise our previous preference. For similar reasons, the recurrence of negative emotions from time to time does not necessarily decrease their intensity. Symmetrically, if during the time-span when they are recurring we experience stronger disappointments, we may not feel quite as bad about the old ones. Still, the recurrence of negative emotions here is a direct incentive to revise our priorities.

Committing Beliefs

The irreducibility of affects to pure effects of combinations between beliefs and desires seems to make it more difficult to understand how emotions could be incentives to revise our beliefs. This is the case even when the influence of emotions on beliefs is documented, at least in terms of the difference between negative and positive moods or emotions (Fiedler, 1990; Fiedler & Bless, 2000; Fiedler & Forgas, 1988; Forgas, 1992, 1995, 2000; Forgas & Bower, 1987; Forgas, Bower, & Krantz, 1984). However, influence can be a pure causal relation, while revision requires some justification. The example of Paul has shown that emotions can be incentives for changing the order of priority between our desires. But changes of priorities without rational justification could lead to epistemic incoherence, and this does not seem compatible with revision of beliefs, which is supposed to restore coherence. Emotional revision cannot be reduced to a pure affective phenomenon without any implication of beliefs. If this were the case, then changing our beliefs about the present fulfillment of our desires would not change the order of our desires.

I would suggest that we have to take into account an often-neglected (maybe because it is too obvious) aspect of beliefs, which is their capacity to monitor and guide our involvement in our activities. Under this aspect, beliefs commit us to choose to do things in one way rather than another: they are “committing beliefs.” They seem to ensure the articulation between the epistemic (purely cognitive beliefs) and the conative sides (desires, motivations). Of course, in order for these committing beliefs to involve us in a dynamics, desire is necessary as a starting factor. However, desire is not sufficient for monitoring, which implies taking into account factual constraints, imposed by external factual dynamics, to which epistemic beliefs are sensitive. Therefore committing beliefs imply the interaction of internal and external dynamics.

Do they not seem to play the same role as affects then? There is a difference: affects are immediate reactions to the impact of external dynamics on internal dynamics. Committing beliefs make us able to select, from among different possible reorientations of our internal dynamics, those that are compatible with our external dynamics. They imply that we consider different possibilities instead of just one reaction and that we make a justified choice, which satisfies the condition of justification of a revision.

What Kind of Order Then for Emotional Revision?

One question still remains. How do we justify a revision of the priorities between desires? Most of the logics of revision assume that revision needs to refer to a fixed order of priorities. For example, AGM revision (Gärdenfors, 1988) assumes a hierarchy of beliefs, ordered according to how “entrenched” they are. Revision consists in cancelling the least entrenched beliefs from among the sets of beliefs that are potentially responsible for the incoherence of the beliefs base once the new discrepant information has to be added to it. Bayesian revision (e.g., Ajzen & Fishbein, 1975; Boden & Berenbaum, 2010; Harlé, Shenoy, & Paulus, 2013; Steunebrink, Dastani, & Meyer, 2012) assumes that before the revision a weight (a probability) has been assigned to each possible state. Possible states are not arranged in an ordinal list, but as subsets of the set of all possible states: a partition. Revision changes the relative weights assigned to the states, but not the partition of the states, even if it assigns zero probability to some states. When we learn information that decreases the probability (the degree of belief) of a possible state, we distribute the difference of probability among the other possible states, proportionally to the previous probability of each of these states. In one step of revision, the rank of a state in the distribution of probability can change but only by keeping the mutual ranking of the other states constant. But strong emotions can cause change of priorities to be spread over the mutual ranking of other states.

We could say that in order to change an order of priorities, we just have to decide to change it. However, this decisionist attitude would prevent us from keeping priorities that guide us from one step of revision to another. It would prohibit any continuity in our motivations as well as any intertemporal justification between different steps and any intertemporal rationality. Fortunately, a (modest and less conservative) justification of changing our priorities is possible if we admit a weaker order. This is what Lehmann et al. (2001) called “pseudo-distance.” Pseudo-distance (d) does not require symmetry: the value of pseudo-distance d from a to b, d(a,b), can be different from the value of pseudo-distance d from b to a, d(b,a). Pseudo-distance is not dependent upon the metric distance between a and b but only upon the number of intermediary things detected between a and b. Pseudo-distance d(c,d), seen from a in a,b,c,d, may be hidden by the closer element b and invisible from a (Schlechta, 2004, pp. 243–245).

Asymmetry may induce changes in the order of priorities. For example, let us consider a triangle, the sides of which are oriented: a → b → c. We can go from a to b, from b to c, and from c to a, but not in the opposite direction. Seeing things from a, the pseudo-distance between a and c (notated a → c) is greater than a → b as you have to pass b in order to get to c; the order of distances is a → c > b → c. From c, in order to get to b, we have to pass a; the order is c → a < c → b. The priorities have been changed, as in the example of Paul.

Paul’s initial work was a restricted part (P) of a collective task. From viewpoint P, his order of priorities was P < I < C, where I represents individual responsibility of a whole task and C represents innovative cooperation. Taking C as a viewpoint, Paul experiences cooperation but in state C feels recurrent negative emotions induced by stress and pressure. C and P become more similar, because both are restrictive situations. While P remains a source of negative feelings, I does not raise recurrent negative emotions. As the relation of order (the pseudo-distance) is not necessarily symmetrical, Paul could now rank P as being nearer to C than to I. His order of priorities is now P < C < I. Such inversion of priorities should not be disqualified as irrational if Paul has no justifications to build any order with better properties. His is the usual human condition: we generally have no means of knowing in advance how we could succeed in defining a unique good order between all our priorities, given our uncertainty about our future recurrent experiences. We can only try to select from our successive revisions the best governing principles. Our highest priorities may resist changes of order, but the results of the revisions of lower priorities may depend on the order in which diverse recurrent emotions will trigger different revisions.

When Easy and Difficult Revisions Are Intertwined

In addition, a revision can be delayed because it is related to a desire of high rank. The rank of the committing beliefs connected with this desire is higher, and the higher this rank is, the more difficult and slower is the revision (the lower this rank, the easier and faster revision is). During the time when the revision is going on, the priorities between our committing beliefs are no longer clearly defined, and this weakness can raise emotion, particularly anxiety related to uncertainty about which commitment has priority. The emotional recurrence of anxiety leads us, in turn, to revise the priority of the difficult revision. Priority may be given to focalization on distracting activities, and this naturally makes the former revision even more difficult.

It is possible that these diverting activities could be easier (and faster) revisions. An easier revision may solve part of the difficulty but not the whole problem, and this leads to suboptimal behaviors. Because the problem linked to the difficult revision has not been solved, anxiety reappears. Its emotional recurrence is itself an incentive for revising the revision of the priority of the easy revision, and that reminds us of the priority of the difficult revision. In turn, once again, the recurrence of the difficulties associated with the difficult revision brings about the revision of its priority. As a result, the priority of easier revisions is reaffirmed. While easier revisions, which may have very tenuous links with the initial problem, might be preferred, they have little chance of providing a solution. As a result, the need for the difficult revision becomes stronger. This revision is still at work, even if recurrently replaced at the conscious level by easier revisions. It is not possible to achieve the revision of either the priority of the difficult revision or the priority of the easier revision, and both are trapped in a cycle of unaccomplished revisions. The activation of an easier revision can mask the difficult revision, but it does not sever the link between them. There is increased likelihood that it will be reactivated over and over again in this cycle, and it can become obsessive. In such a case, the anxiety raised by the difficult revision diffuses over the easier revision. But as anxiety is not justified by the latter revision, it seems neurotic (for other combinations of difficult and easier revisions, see Livet, 2002).

Self-deception may be understood as an effect of these problems of entangled difficult and easier revisions. If someone has to make a big revision of his or her usual behavior, for example, to stop smoking, an easier revision may be to focus—self-deceptively—on any argumentation that decreases the evidence about the correlation between smoking and cancer.

These problems are related in part to the different temporal scales of the different revisions. In the short term, the diversion towards the easier revision can be done first because the easier revision is faster than the difficult revision. In the medium term, the anxiety related to the difficult revision diffuses over the easier revision as soon as the two become linked. Gilbert and Ebert (2002) observed another intertemporal linkage: participants who had to make a decision preferred to have the possibility of changing their minds in the future. As choice implies revision, the participants preferred to ease the anxiety related to immediate revision, even though their future choice would add to its anxiety the regret that stems from the feeling that things could already have been settled.

The so-called “transmutation” of emotions (Elster, 1999) is also an effect of easier revisions. For example, people who envy the deserved promotion of a colleague, which implies a difficult revision of their expectancies, prefer to become more receptive to the information that the other person did not deserve the promotion. In so doing, they avoid being ashamed of their envy (on shame, see Gratz, Rosenthal, Tull, Lejuez, & Gunderson, 2010). There is no real “transmutation,” as envy is still there. The diversion created by indignation only separates the envy from the difficult revision.

Taking Diversity Into Account

It could be disputed that while this theory holds true for anxiety, which is related to uncertain revisions (Gasper & Clore, 1998; for interpretive biases, see E. J. Wilson, MacLeod, Mathews, & Rutherford, 2006), it cannot take into account the diversity of emotional phenomena. However, the analysis of the revision of committing beliefs process, combined with the definition of affects as an interaction between an internal and an external dynamics, makes it possible to reinterpret some of the evidence of this diversity. The difference between arousal and valence (e.g., Nicolle & Goel, 2013) follows from the definition of affects. Arousal is related to the magnitude of the difference between the two dynamics that interact in affects. Positive valence implies that external dynamics enhances the internal dynamics or stops a decreasing dynamics. Negative valence implies that it impedes a stable or increasing dynamics or accelerates a decreasing dynamics. In an “integral” emotion (raised by the features of its target), the intentionality of the external dynamics is centrally relevant for the internal dynamics. In an incidental emotion (raised by stimuli not directly related to the target), it is relevant but only for other peripheral internal dynamics and motivations (Blanchette & Richards, 2010; Pham, 2007; A. Richards, Blanchette, & Munjiza, 2007).

The paradox of affect-infusion implies that while incidental affects have little influence on quasiautomatic emotional reactions, they have a stronger influence on more elaborated cognitive processing (Forgas, 1992, 2000, 2011). Each step of the latter may reveal ramifications (possibly revisable ones), which divert us from only focalizing the final goal and make us accessible to peripheral information (see also Banich et al., 2009, on the relation between emotions and cognitive control).

Positive moods make us not only more sensitive to stereotypes and less attentive to the details, but also more creative (Bless et al., 1996; Fredrickson & Joinder, 2002; Gabble & Harmon-Jones, 2010; Isbell, Burns, & Haar, 2005; Koch & Forgas, 2012). Negative moods have the opposite effects (Bless et al., 1990; Fiedler & Bless, 2000; Fiedler & Forgas, 1988; Fiedler, Nickel, Muehlfried, & Unkelbach, 2001; Forgas, 1992, 1995, 2000, 2009; Fredrickson, Cohn, Coffey, Pek, & Finkel, 2008). Moods imply a rather stable orientation of internal dynamics. People in a positive mood are less sensitive to discrepancies with their expectancies. They follow the track of their usual expectancies because their increasing dynamics has a stronger inertia that explains stereotypy. Moreover, resistance to the interpretation of discrepancies as contrary to expectancies explains open-mindedness and creativity. People in a sad mood are not only oversensitive to discrepancies and bad news, but also to the discrepancy of good news; they are in quest of more informative details (Ambady & Gray, 2002; Isen, Nygren, & Ashby, 1988; Mathews & MacLeod, 2002).

If anger and sadness are negative emotions, why do angry people take risks in the same way that happy people do (Constans & Mathews, 1993; Kahneman & Tversky, 1984; Lerner & Keltner, 2001; Mittal & Ross, 1998; Raghunathan & Corfman, 2004; Raghunathan & Pham, 1999; Raghunathan, Pham, & Corfman, 2006)? The strong internal dynamics of angry people encounters an external opposition (Dunn & Schweitzer, 2005; Siemer, 2001; Siemer et al., 2007) that they try to overcome. If this is not possible, the strength of the dynamics has no outlet other than the expression of anger. Impotent recurrent anger can lead to bitterness (and depression, see Besharat, Nia, & Farahani, 2013; Riley, Treiber, & Woods, 1989). The internal dynamics of happy people is strong enough to pass over discrepancies. The dynamics in both of these cases are assumed to exceed the opposed discrepancies, but because there may be little grounds for such an assumption, both are at risk. The internal dynamics of sad people is weaker, so they take fewer risks.

Is it possible to take into account a variety of emotions, such as anxiety, disgust, amusement, admiration, contempt, and guilt? Anxiety is related to the prospect of possible revisions. Disgust differs from fear (in which the external dynamics tends to destroy or invert our internal dynamics, leading to three possible reactions, namely stopping our dynamics, taking flight, or strengthening our dynamics and becoming aggressive) in that the problem of the repulsive external dynamics relating to disgust is dealt with by simply avoiding the repulsive cause. Amusement enhances our dynamics only by comparison with the dynamics of other people confronted with discrepancies, which we could easily reduce or which are irrelevant for our main goals (ironic stance is amusement about one’s own difficulties). Admiration (in its general meaning) is a rather contemplative reaction to external dynamics that reinforce some of our expectancies but with which we have no present program of interaction. Contempt is a more specific reaction to dynamics that undermine our compliance to social norms (compliance is a rather complex interaction between dynamics and the subsequent revisions). Guilt is the emotion raised by acknowledging that we have been the source of a negative dynamics for another person, while also acknowledging that this decreases our own present dynamics.

Repeated Emotions: Different Temporalities

Not only are emotions diverse, so too are the temporalities of their repetitions. Repetitions at short term do not lead to revision proper, but rather to habituation or sensitization (Ben-Haim, Mama, Icht, & Algom, 2014; Compton et al., 2003; Davidson, Scherer, & Goldsmith, 2009; Fiedler & Bless, 2000; Fiedler et al., 2001; Kalisch, 2009; Rhudy, Bartley, & Williams, 2010; Tryon, 2005), which may either increase or decrease the impact on our internal dynamics. Repetition at medium term leads to what can be called “accustomization.” The effects of accustomization last longer than those of habituation and sensitization. Accustomization does not exclude sensitivity to opposite information. Fredrickson et al. (2008) noted that, in the medium term, even if the broad-minded coping that is associated with happiness increased positive affects, the negative affects coming from other sources during this positive mood were not neglected. In both cases, after a while, our internal dynamics recovers its usual trajectory. Both Fredrickson and Joinder (2002) and Kashdan, Young, and McKnight (2012) analyzed long-term effects with similar conclusions.

Emotional recurrence is a long-term effect (longer intervals than those leading to habituation and accustomization, see Reisberg, 2013, pp. 585–599). The time interval between two similar emotions is sufficient for the internal dynamics to recover its usual process. In addition, the memory of the recurrence is accessible. The impact of the external dynamics of the present reoccurrence is recognized as recurrent. As such, it impacts not only the present internal dynamics, but also its potentiality to reoccur and its capacity to return to its usual process. This is why emotional recurrence is an incentive for revising the expectancies and priorities between the desires linked with the usual internal dynamics.

Some incentives can have bad results, for example recurrence of dissatisfaction can change sadness into resignation, and if the rank of the related desires is revised to a very low level, depression sets in (Ehring, Fischer, Schnülle, Bösterling, & Tuschen-Caffier, 2008). Rumination at short intervals on a past emotional event predicts depressive disorders and anxiety (Nolen-Oeksema, 2000). It repeatedly evokes the traumatic impact of the external dynamics without revision. The exception to this is when rumination is analytic, as the cognitive abilities needed for revision are activated (e.g., De Lissnyder et al., 2012; Morgan & Banerjee, 2008; Most, Chun, Widders, & Zald, 2005). Rumination, because it repeatedly focuses in the medium term on the same failure of the subject’s expectancies and it blocks others expectancies and dynamics, contrasts with the emotional recurrence that leaves space for a reemergence of the previous dynamics and for other expectancies.

Emotional Revision and Values

A traumatic (or ecstatic) event can be seen as the concentration in one event of the incentive power of a long emotional recurrence. However, it does not necessarily trigger a revision of the priorities between our desires. This seems to be because the attacked priorities are too deeply entrenched to be revised. If our life has been put in danger, its priority is still fundamental. This justifies differentiating between the general category of preferred desires and the more specific one of values. Both are related to priorities, as “value” could be a summary for “having a rank in an order of values.” It would be tempting to define real values as deeply entrenched desires, resistant to revision. But in the example of Paul, “individual responsibility” and “innovative cooperation” are values. Nevertheless their order of priority has been revised. Can any value be revised?

We can use the differences between the temporalities of the repetition of emotions and their different effects on change or revision as filters for distinguishing values and their different kinds inside the general category of preferences. Values seem to be more capable than temporary desires of resisting temporary changes. As habituation and sensitization are changes of preferences in the short term, values would not be a suitable term here. A sign of real values could be their resistance to accustomization. Values can be seen as priorities that can reappear even after we have become accustomed to different temporary preferences. However, as shown in the story of Paul, some values are not immune to the revision of their priority. The recurrence of similar emotions after a lapse of time sufficiently long for the previous dynamics related to these values to reappear could undermine their capacity of reemergence. Only fundamental values resist revision by emotional recurrence. As changes of superficial preferences do not mobilize emotional revision (habituation and accustomization are sufficient), and fundamental desires and values resist it, its impact is effective mainly on values of medium levels.

The Paradox of Emotional Sharing

Does this limited extension of the power of emotional revision shed light on the “paradox” of emotional sharing, as described by Rimé, Luminet, Philippot, and others? These authors (Brans, Rimé, van Mechelen, & Verduyn, 2014; Di Schiena, Luminet, & Philippot, 2011; Luminet, Bouts, et al., 2000; Luminet, Zech, et al., 2000; Rimé, 2009; Rimé, Finkenauer, Luminet, Zech, & Philippot, 1998; Rimé et al., 2011; Rimé, Philippot, Boca, & Mesquita, 1992) have shown that people share their emotions with their partners, particularly about traumatic events. The paradox is that this social sharing does not induce emotional recovery, although people feel relief after sharing their emotions.

Emotional recovery is usually defined (see, for e.g., Stroebe, Schut, & Stroebe, 2006) as the progressive decrease in emotional arousal or intensity when people recall the traumatic event. For Rimé (2009) and Rimé et al. (2011), emotional recovery required the cognitive treatment of the emotional experience, the reorganization of motives, the reframing and reappraisal of the emotional event and its situation (Gross, 1999; Keng, Robins, Smoski, Dagenbach, & Leary, 2013; Kross, Davidson, Weber, & Ochsner, 2009; Moser, Most, & Simons, 2010; Ray, Gross, & Wilhelm, 2008; J. M. Richards & Gross, 1999; Siemer et al., 2007; Whittle, Walker, Medd, & Mort, 2012). These changes bear a striking resemblance to those observed in emotional revision.

Traumatic events undermine fundamental values. These values are related to the very internal dynamics that have the capacity to reemerge even when they have been recurrently undermined, and are therefore resistant to emotional revision. For example, Rimé (2009) noted that bereaved people might not revise the committing beliefs associated with the expectancy of interacting with a recently deceased parent. This explanation could be challenged on the grounds of its circularity: the difficulty of revision is supposed to indicate the high rank of desires and expectancies, and this high rank is assumed to explain how difficult the revision of their associated committing beliefs is. This circularity is broken, however, because (a) the fundamental priorities that we assume to be implied in resisting revision of the violated expectancies can be detected independently of the specific resisting expectancies (the bereaved still give high priority to their attachment to their remaining nearest relatives), (b) revision of committing beliefs can be detected independently of emotional recovery (Blanchette, Richards, Melnyk, & Lavda, 2007; Rimé, 2009).

Why do we like to share strong emotions with others and why do we feel it useful? When very high priorities are at stake in an emotional event, it can be difficult to know how to calibrate the revision. In emotional sharing, people have to refer to social scenarios that help them to define this calibration. Rimé (2009) has observed that we do not share emotions socially when their intensity is either too low or too high. Emotions that are too weak are assumed not to give rise to compassion in other people. Emotions that are too strong are difficult to share. This may be an effect of the relation between high priorities and resistance to revision (see Fredrickson, Tugade, Waugh, & Larkin, 2003). Victims’ emotions that are too intense make other people suppose that they might also have to revise some of the fundamental priorities that they share with the victims. Other people may be prone to sharing our emotions, but they are reluctant to share revisions of their own basic expectancies. Nevertheless, when emotional sharing is accepted, it is a way of confirming how robust these priorities are. As other people share our emotions, we are entitled to restart our undermined dynamics. They also help us to socially calibrate these dynamics and to assign to them new and more adapted objectives. These two effects can explain why people feel comforted by such social sharing.

Why does emotional sharing not ensure emotional recovery? On the one hand, if our expressed emotion is too weak (because we try to socially calibrate it), our partners might have difficulty feeling compassion. On the other hand, emotional sharing implies reliving the emotion, which triggers the revival of the dynamics related to the attacked priorities, reinforcing the intensity of emotion. In addition, as other people do not feel such intense emotions (they can maintain their basic dynamics without strong revision, in contrast to us), the suspicion that they cannot understand our pain enhances our negative feelings. In this respect, social calibration and revival are opposite one another, while in another respect social calibration allows the adaptive restart of our basic dynamics. While the resources for adaptive and revised behaviors are also still possible causes of resistance to changing the committing beliefs of our fundamental dynamics, emotional recovery is not possible.

Conclusion

I hope I have shown that emotional recurrence is an incentive for revising our committing beliefs and that this revising power implies that affects have to be understood as the interactions of external dynamics with our internal dynamics. Different effects of the different temporalities of the repetition of similar emotions are useful indicators of the different priorities that we attached to different values. The order that is proper to emotional revision has specific properties, and these are required for changes in the order of our preferences to be possible. Its flexibility shows not only the existence but also the limits of emotional rationality.